6/4/2022

Mission

Mission

  • 172 MVP candidates from 2010-2022
  • With 40 total variables
##     mvppts ptsmax share games  mpg  ppg rbpg astpg
## 113      2   1010 0.002    62 33.4 25.1  8.3   4.8
## 82    1198   1300 0.922    80 32.7 23.8  4.3   7.7
## 75      26   1250 0.021    69 36.2 23.2 11.1   2.6
## 36     889   1210 0.735    66 38.6 28.0  8.0   3.5
## 25     522   1210 0.431    79 38.8 26.7  7.5   7.0
  • And 34 more variables…

Sentiment Analysis

Sentiment Analysis

Example Text Sentiment Score (VADER[2])
According to Dirk Nowitzki, reporters should pay no attention to Jason Terry after games -0.382
[Draymond Green in 2010] “the world is flat fool” -0.440
Stephen Curry is 5th All-Time in career True Shooting Percentage at 62.0%, and 1st All-Time among guards 0.421
[Murphy] Serge Ibaka interrupts DeRozan’s scrum to ask what he got the team for Christmas. Told nothing, Ibaka says he’s going to go play with John Wall. 0.026

Modeling

Modeling

Important Features

Important Features

Results

  • Model was able to predict the MVP winner with 77% accuracy (2010-2022)
  • With average deviation from actual rank of 0.41

Predictions for 2022 Season

Player Rank Predicted Rank Sentiment Score
Nikola Jokic 1 1✅ 0.94
Joel Embiid 2 2✅ -0.22
Giannis Antetokounmpo 3 3✅ 1.00

Future

  • Test out model midseason
  • Add text sentiment analysis of famous critics like Stephen A Smith
  • Use comments instead of headlines
  • Comedy scores
  • Account for jargon

Reflection

  • Learned the arduous task of gathering data
  • How we different models can be used

Resources Used For Presentaiton